Re: [tcpm] How to improve TCP efficiency in a large RTT context?

[Mark Allman <[e-mail]>]

> · Is there room to increase even more the Initial Window? Several
> major players already use IW=32 or higher

I'll just flog this again: http://www.icir.org/mallman/pubs/LAJW07/

:)

(I chaired TCPM when that came out.  The IETF subsequently wised up
and found better people.  Lars was the TSV AD at the time.  His
comment on the above when it was presented at the workshop was
something along the lines of: can we really have a TCPM chair who
would propose an arbitrary IW?!  I think he meant it as a joke.  I
took it as a compliment. :) )

But, in all seriousness, we don't need to specify the IW.  Or, how
it gets set.  As long as the rest of CC is in place, there are
consequences for being too aggressive that will keep things
reasonable.  IMHO.  Not to mention that the actual impact of lousy
choices isn't all that bad.  And, of course, if we never had to
discuss the IW again, that'd be cool, too. :)

allman


--
http://www.icir.org/mallman/